The People's Republic of Walmart

by Leigh Phillips

  The then-29-year-old head of the Chilean Production Development Corporation, Fernando Flores, responsible for the management of coordination between nationalized companies and the state, had been impressed with the prolific writings on management cybernetics of a British operations research scientist and management consultant named Stafford Beer. Flores had studied industrial engineering at the Catholic University, but in doing so, he had also trained in operations research, that branch of applied mathematics in search of optimal solutions to complex decision-making problems. It’s a salmagundi of a discipline, drawing on modeling, statistical analysis, industrial engineering, econometrics, operations management, decision science, computer science, information theory, and even psychology. In the course of his studies and early work for the Chilean railways, Flores had come across Beer’s texts on cybernetics. While Beer’s work, for which he had gained a substantial international reputation, focused on more efficient management techniques, according to Medina’s interviews with Flores, the latter was captivated by how the “connective, philosophical foundation” of Beer’s management cybernetics could serve Allende’s vision of an anti-bureaucratic democratic socialism in which workers participated in management and that would defend individual civil liberties. Management cybernetics, Flores reasoned, could assist the young government in “herding the cats” of the public and worker-managed sectors.

  The term “cybernetics” today has something of a naively techno-utopian aura, or even a body-horror, dystopic dread about it. But at its fundament, the field of cybernetics simply investigates how different systems—biological, mechanical, social—adaptively manage communication, decision making and action. The first edition of Beer’s 1959 book on the subject, Cybernetics and Management, does not even make reference to computers, and, as Medina is keen to stress, Beer himself was an intransigent critic of how business and government deployed computers. Cybernetics is not management by algorithm. It is not digital Taylorism.

  During World War II, MIT mathematician Norbert Wiener and his engineering colleague Julian Bigelow were tasked with developing ways of improving the targeting of enemy aircraft. Following consultations with an early neuropsychologist, the two developed an apparatus that automatically helped the human gunner to correct their aim through what they called feedback, a circular method of control through which the rules governing a process are modified in response to their results or effects. Today, this may seem obvious (and its very obviousness is likely a product of how influential cybernetic notions have become in our culture; this is where the word “feedback” comes from), but at the time, this was a revelation wherein linear, “if this, then that” control systems dominated. As Richard Barbrook recounts in his 2007 history of the dawn of the computer age, Imaginary Futures, despite the military engineering origins of the field, Wiener would go on to be radicalized by the Cold War and the arms race, not only declaring that scientists had a responsibility to refuse to participate in military research, but asserting the need for a socialist interpretation of cybernetics. “Large corporations depended upon a specialist caste of bureaucrats to run their organisations,” Barbrook notes. “They ran the managerial ‘Panopticon’ which ensured that employees obeyed the orders imposed from above. They supervised the financing, manufacture, marketing and distribution of the corporation’s products.” Wiener, and later Beer, on the other hand, conceived of cybernetics as a mechanism of domination avoidance: a major challenge that the managers of any sufficiently complex system face, according to Beer, is that such systems are “indescribable in detail.”

  Echoing this concern, three years before the Prague Spring uprising of workers and students was crushed by Soviet tanks in 1968, two Czechoslovak authors, Oldrˇich Kýn and Pavel Pelikán, published Kybernetika v Ekonomii, a book that challenged the top-down central planning system. In it, they focused on the key role of accurate information in the coordination of economic activities, whether via the market or through planning, arguing that the human capacity to receive and process information is inherently limited. A high degree of centralizing hierarchy requires that the top-level decision makers have a large information-processing capability. At the same time, in addition to the poor quality of decision making resulting from the inability of an individual or even a small group of humans to process more than a certain amount of information, overcentralization can also result in the costs of transmitting and processing information being “many times higher than the most pessimistic estimates of loss that could occur with an effective reduction of information and a decentralization of a large part of the decision-making.” Instead, Kýn and Pelikán proposed that the amount of information be gradually reduced along the hierarchy, with each place in the hierarchy enjoying a certain degree of freedom for independent decisions: “Not all the information collected below can arrive at the highest places. The problem, of course, is how to reduce information without losing what is essential for making decisions.”

  Conversely, as Beer was aware, too much decentralization and autonomy could produce chaotic results that undermine the well-being of the system as a whole, producing either debilitating overproduction or shortages. Thus his model aimed to promote a maximum of self-organization among component parts via redundant, lateral, multi-node communication networks, while retaining some channels of vertical control to maintain systemic stability and long-term planning. Instead of the abstract dichotomy of centralization versus decentralization, he asked: What is the maximum degree of decentralization that still permits the system to flourish?

  Allende was attracted to the idea of rationally directed industry, and upon Flores’s recommendation, Beer was hired to advise the government. Beer, for his part, frustrated at only ever seeing partial implementation of his ideas by the firms he advised, was attracted to the possibility of putting his full vision into practice, and on a much wider scale than he had yet attempted.

  And that vision would involve the linking-up of a realtime communication network, connecting factory floor to factory floor, and upward to the State Development Corporation (CORFO), rapidly dispatching data both laterally and vertically and thus allowing quick responses at all points in the system to changing conditions. The data collected would also be crunched by a mainframe computer to produce statistical projections about likely future economic behavior. In addition, the system would involve a computer simulation of the Chilean economy as a whole, which Beer and his colleagues termed “CHECO” (CHilean ECOnomic simulator). However, upon his first visit to Chile, Beer was confronted with the reality of the country’s limited computer resources: just four low- to mid-range mainframes owned by the National Computer Corporation (ECOM), which were already largely locked up with other tasks. At most, ECOM could offer processing time on one such device, an IBM 360/50. As Medina puts it, Beer would be building a computer network of one computer.

  But the key was the network, not the type of machine doing the networking. And so as a solution, Beer suggested connecting to the single IBM mainframe a communications network of telex machines—those very ’70s automatic typewriter–looking devices you see in All The President’s Men, direct descendants of the telegraph system and first manufactured in the 1930s—which were common enough in Chile and at the time were even more reliable than telephones. Initially, Beer thought he was working on a project to develop a more accountable, more responsive communications and control system between government-appointed factory managers, or “interventors” to use the Chilean terminology of the time, and CORFO. He envisaged that the interventors at each enterprise would use the telex machines to transmit production data to the telex machine at the National Computer Corporation. Computer operators there would then translate this information into punch cards that would be fed into the mainframe, which would in turn use statistical software to compare current data with past performance, seeking anomalies. If such an anomaly were discovered, the operators would be notified and they would then notify both the interventor concerned and CORFO. CORFO
would then give the interventor a brief period to sort out the anomaly on their own, offering the enterprise a certain degree of autonomy from higher decision making while also insulating those government decision makers from what could otherwise be a tsunami of data by transmitting only what was crucial. Only if the interventor could not sort out the problem would CORFO step in. In this way, instead of all production decisions being made in a centralized top-down fashion, there would be an iterative “roll-up” process, as Beer described it, with policies transmitted downward to factories and factories’ needs transmitted upward to government, continually adapting to new conditions. Beer, a severe critic of Soviet bureaucracy, also believed that the statistical comparisons produced centrally would reduce the ability of factory managers to produce false production figures, as happened in the USSR, and enable much-faster discovery of bottlenecks and other problems. The aim was real-time economic control—in this period a staggering ambition, socialist or otherwise—or as close to it as possible. Up until this point, Chile’s conventional economic reporting methods involved extensive, lengthy printed documents detailing information collected on a monthly or even yearly basis.

  Paul Cockshott, the computer scientist whom we met earlier, who has written extensively on the possibility of post-capitalist planning aided by contemporary processing power, is a big admirer of Cybersyn: “The big advance with Stafford Beer’s experiments with Cybersyn was that it was designed to be a real-time system rather than a system which, as the Soviets had tried, was essentially a batch system in which you made decisions every five years.”

  Allende, too, as soon as he was familiar with precisely how the system worked, pushed Beer further to expand its “decentralising, worker-participative and anti-bureaucratic” possibilities. Allende’s desire that Proyecto Synco not be a technocratic answer to economic planning along Soviet lines, but a tool in the hands of workers on the shop floor to engage in decision making themselves, impressed Beer and hinted at a much-wider application of the system than just the nationalized sector. But then, Beer was also being radicalized by events in the country beyond the consulting work he was performing for CORFO. Allende was pushing at an open door.

  Even before the election of Allende’s six-party Popular Unity coalition government, the United States had spent millions on propaganda efforts against the Left and to support the Christian Democrats. Upon the nationalization of the copper industry (even with the unanimous support of the opposition Christian Democrats), Chile’s primary export, the United States cut off credits, and the multinational companies that had been the owners of the mines worked to block exports. Factory and land owners took to the courts to try to block reforms, and sections of the Right openly called for a military coup, an option supported by the CIA. While substantial wage increases for manual and white-collar workers had initially slashed unemployment and contributed to strong economic growth of 8 percent a year, this de facto blockade soon crippled the economy and limited the availability of consumer items. With wage increases chasing fewer items, shortages and crippling inflation appeared, in turn provoking accusations of middle-class hoarding. Allende’s Popular Unity government—very much believed by the working class to be their government—was being threatened internationally and domestically. The workers and peasants were radicalizing; society as a whole was sharply polarizing.

  The circumstances of a government under threat forced Beer’s team to work under a crash schedule. The project was challenged on a number of fronts that were not eased by the acceleration of the timetable, but the difficulties were less technological than they were social. Operations research scientists had to perform studies of every nationalized company and establish which production indicators the software would need to track and which ones it should ignore. This was no simple task, even for a simplified model that was intended not to represent the full complexity of the Chilean economy, but simply to uncover the key factors that had the biggest impact on outputs. Nevertheless, the CHECO model was to go beyond production factors—productivity and demand—to consider the currency supply—investment and inflation. But the team was having difficulty simply getting hold of the necessary information to test the model. Mining data was two years old. Agricultural data was sparse. In some enterprises, advanced information collection processes did not even exist. In the end, while CHECO was able to run experimental models exploring inflation, foreign exchange and national income, as well as simplified models of the whole economy and of a handful of sectors, the team viewed these efforts only as a testing ground, not to be used to develop policy.

  In addition, for all of Beer and Flores’s desire and Allende’s insistence that the project achieve a participative, decentralizing and anti-bureaucratic system, the role of workers on the factory floor was sometimes negligible, with Cybersyn engineers tending to speak first to enterprise upper management, then to middle management, and then finally to factory production engineers. Medina’s history of the project is careful not to romanticize the results. The engineers did consult with workers’ committees, but not on a regular basis. On top of this, to be able to model individual factories, they needed postsecondary training in operations research, and Chile at this time had a very limited pool of graduates who had been so trained. The team faced resistance from factory managers, whose class position made them less sympathetic to the project, or who simply did not understand what its purpose was. Despite direction to factory engineers that they work with workers’ committees, again class divisions posed a barrier: engineers were instead condescending to workers and preferred talking to management. Medina, in her research, found very little evidence that rank-and-file workers played much of a role in shaping the modeling process.

  But one can also imagine the same system being used in a very different way, arming instead of disarming workers. Indeed, even in embryo form, the Cybersyn communications network helped groups of workers to self-organize production and distribution during what would otherwise have been a crippling trucking strike, mounted by conservative business interests and backed by the CIA, in 1972. In so doing, it offered the struggling Allende administration a brief stay of execution.

  Cyber Strikebreaking

  It was during the strike that Cybersyn came into its own. The network could allow the government to secure immediate information on where scarcities were most extreme and where drivers not participating in the boycott were located, and to mobilize or redirect its own transport assets in order to keep goods moving. But this was not a simply a top-down operation, directed from La Moneda Palace by the president and his ministers. The strike had forced public sector operations that were near to each other to work together in “cordónes industriales”—literally, “industrial belts”—in order to coordinate the flow of raw materials and manufactured products. The cordónes in turn worked with local community organizations, such as mothers’ groups, to assist with distribution. The autonomous operation of these cordónes mirrored forms of spontaneous worker and community self-direction that appear to pop up regularly during times of revolutionary upheaval, or otherwise at times of crisis or natural disaster, whether we call them “councils,” “comités d’entreprises” (France), “soviets” (Russia), “szovjeteket” (Hungary) or “shorai” (Iran). Liberal commentator Rebecca Solnit describes in her social history of the extraordinary communities that emerge at such extreme moments, A Paradise Built in Hell, how, far from the chaotic, Hobbesian war of all against all of elite imagination, it is calm, determined organization that on the whole prevails. She repeatedly discovered how remarks by those attempting to survive through earthquakes, great fires, epidemics, floods and even terrorist attacks that despite the horrors experienced, reflect how truly alive, full of common purpose and even joyful they felt. It is no wonder that a rich, long-lived stream of libertarian socialist thought, running through the writings of the likes of Rosa Luxemburg, Anton Pannekoek, and Paul Mattick emphasizes such organization, such “councils,” as the foundation of the free society t
hey wish to build. The great challenge is the scaling-up of such democratic, market-less organization. This is the distilled version of the economic calculation debate: relatively flat hierarchies seem perfectly capable of democratically coordinating production and distribution for a limited number of goods and services, for a small number of people and over a limited geography. But how could the myriad products needed by a modern, national (or even global) economy—with its complex web of crisscrossing supply chains, thousands of enterprises and millions of inhabitants (billions, if we consider the global case)—be produced without vast, metastasizing and inefficient bureaucracies? How are the interests of the local production node integrated harmoniously with the interests of society as a whole? What may be in the interest of a local enterprise may not be in the interest of the country.

  What happened in Chile in October of 1972 may not be the definitive answer to these questions, but it hints at some possibilities.

  On October 15, Flores suggested to the director of the CHECO project that they apply what they had learned from their experimentation to battling the strike. They set up a central command center in the presidential palace, connected via the telex machines to a series of specialized operational units focusing on different key sectors: transport, industry, energy, banking, supply of goods and so on. This network allowed the government to receive minute-by-minute status updates directly from locations across the country, and then just as quickly to respond, sending orders down through the same network. A team at the palace analyzed the data coming in and collated them into reports, upon which government leaders depended to make decisions. If one factory was short of fuel, spare parts, raw materials or other resources, this data flowed through the network to another enterprise that could help. Information was also shared on which roads were clear of oppositionists, allowing the trucks that remained under public control to redirect themselves and avoid blockades. Medina notes how some historians emphasize, instead, the role of popular mobilization from below in breaking the strike, but she argues this is an unnecessary dichotomy. While it did not eliminate vertical hierarchy, the network did connect the government command center to the horizontal activities on the ground. Medina writes: “The network offered a communications infrastructure to link the revolution from above, led by Allende, to the revolution from below, led by Chilean workers and members of grassroots organizations, and helped coordinate the activities of both in a time of crisis.” She argues that Cybersyn simply faded into the background, “as infrastructure often does.” The system did not tell the workers what to do; the workers and their representatives in government simply used the system as a tool to aid them in doing what they wanted to do.